$\tau$-tilting graphs and quotients

Abstract.

We investigate $\tau$-tilting graphs of algebras and their quotient algebras, and obtain a sufficient condition for the connectivity of $\tau$-tilting graphs to be maintained in quotient algebras. It is worth pointing out that $g$-tame algebras satisfy this condition. As a consequence, we newly obtain a large class of algebras whose $\tau$-tilting graphs are connected, including in particular the quotient algebras of skew-gentle algebras and quasi-tilted algebras of tame type.

1. Introduction

An important combinatorial invariant of a cluster algebra is its exchange graph. The vertices of this graph correspond to the seeds, and the edges connect the seeds related by a single mutation. A cluster algebra is of finite type if its exchange graph is finite, that is, it has finitely many distinct seeds. Cluster algebras of finite type were classified in [FZ03a]: they correspond to finite root systems. Moreover, the exchange graph of a cluster algebra of finite type can be realized as the 1-skeleton of the generalized associahedron, or Stasheff’s polytope [FZ03b]. Through the categorifications of cluster algebras using representation theory one obtains a whole variety of exchange graphs associated with a finite dimensional algebra or a differential graded (dg) algebra concentrated in non-positive degrees. These constructions come from variations of the tilting theory, the vertices of the obtained exchange graph being support $\tau$-tilting modules, torsion pairs, silting objects and so on. For a reasonably complete discussion of the history of abstract exchange graphs stemming from representation theory, see the introduction in [BY13].

This note continues our study on (abstract) exchange graphs in [FGLZ23]. In that paper, along with Zhou, we used $\tau$-reduction [J15] to prove that the exchange graph of support $\tau$-tilting modules over a finite dimensional gentle algebra is connected and has the so-called reachable-in-face property (the terminology will be recalled later). In the current paper, we first observe that if a finite-dimensional algebra has the reachable-in-face property, then every $\tau$-reduction also has the reachable-in-face property, and the connectivity of the $\tau$-tilting graph is also preserved under $\tau$-reduction (see Theorem 3.1). With this as a starting point, we study more general quotient algebras than $\tau$-reductions. Proposition 3.4 gives a sufficient condition for the connectivity of $\tau$-tilting graphs to be maintained in quotient algebras. Using this, Theorem 3.6 proves that for any $g$-tame algebra, if its $\tau$-tilting graph is connected, then the $\tau$-tilting graph of any of its quotient algebras is also connected. As a consequence, we newly obtain a large class of algebras whose $\tau$-tilting graphs are connected, including in particular the quotient algebras of skew-gentle algebras and quasi-tilted algebras of tame type.

Convention

Throughout this paper, let $k$ be an algebraically closed field. By a finite dimensional algebra, we always mean a basic finite dimensional algebra over $k$. For a finitely generated right module $M$ of a finite dimensional algebra $A$, we denote by $|M|$ the number of pairwise non-isomorphic indecomposable direct summands of $M$.

2. Preliminary

2.1. (Support) $\tau$-tilting graphs

Let $A$ be a finite dimensional $k$-algebra. Denote by $\operatorname{\mathsf{mod}}\nolimits A$ the category of finitely generated right $A$-modules. Let $\tau$ be the Auslander-Reiten translation of $\operatorname{\mathsf{mod}}\nolimits A$. Recall that a module $M\in\operatorname{\mathsf{mod}}\nolimits A$ is $\tau$-rigid if $\operatorname{\mathsf{Hom}}_{A}(M,\tau M)=0$. A $\tau$-rigid module $M$ is $\tau$-tilting if $|M|=|A|$. A $\tau$-rigid pair is a pair $(M,P)$ with $M\in\operatorname{\mathsf{mod}}\nolimits A$ and $P$ a finitely generated projective right $A$-module, such that $M$ is $\tau$-rigid and $\operatorname{\mathsf{Hom}}_{A}(P,M)=0$. A $\tau$-rigid pair $(M,P)$ is a $\tau$-tilting pair provided that $|M|+|P|=|A|$. In this case, $M$ is a support $\tau$-tilting $A$-module and $P$ is uniquely determined by $M$ provided that $P$ is basic. In the following, we always identify basic support $\tau$-tilting modules with basic $\tau$-tilting pairs. Denote by $\operatorname{\mathsf{\tau-tilt}}A$ the set of all basic $\tau$-tilting pairs of $A$.

Let $(M,P)$ and $(N,Q)$ be two $\tau$-rigid pairs, we say that $(N,Q)$ is a direct summand of $(M,P)$ if $N$ and $Q$ are direct summands of $M$ and $P$ respectively. A $\tau$-rigid pair $(M,P)$ is indecomposable if $|M|+|P|=1$. In particular, each $\tau$-tilting pair has $|A|$ non-isomorphic indecomposable direct summands. Let $(M,P)$ be a basic $\tau$-rigid pair such that $|M|+|P|=|A|-1$. It has been proved in [AIR14] that there exist exactly two non-isomorphic basic $\tau$-tilting pairs $(M_{1},P_{1})$ and $(M_{2},P_{2})$ such that $(M,P)$ is a direct summand of $(M_{i},P_{i})$ for $i=1,2$. Clearly, $(M_{1},P_{1})$ and $(M_{2},P_{2})$ differ exactly in one indecomposable direct summand, say $(N,Q)$. In this case, $(M_{1},P_{1})$ is called the mutation of $(M_{2},P_{2})$ at $(N,Q)$ and versa.

Definition 2.1.

The $\tau$-tilting graph $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ has vertex set indexed by the isomorphism classes of basic support $\tau$-tilting $A$-modules, and any two basic support $\tau$-tilting modules are connected by an edge if and only if they are mutations of each other.

Let $(L,R)$ be a basic $\tau$-rigid pair. Denote by $\tau$-tilt${}_{(L,R)}A$ the set of $\tau$-tilting $A$-pairs containing $(L,R)$ as a direct summand. We denote by $\mathcal{H}_{(L,R)}(\operatorname{\mathsf{\tau-tilt}}A)$ the full subgraph of $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ consisting of vertices which admit $(L,R)$ as a direct summand.

Definition 2.2.

We say that the $\tau$-tilting graph $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ has reachable-in-face property if for any basic $\tau$-tilting pairs $(M,P)$ and $(N,Q)$ such that there is path in $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ from $(M,P)$ to $(N,Q)$, then for any common direct summand $(L,R)$ of $(M,P)$ and $(N,Q)$, there is a path in $\mathcal{H}_{(L,R)}(\operatorname{\mathsf{\tau-tilt}}A)$ linked $(M,P)$ and $(N,Q)$. In this case, we also say that $A$ has the reachable-in-face property.

The following is useful for describing the subgraph $\mathcal{H}_{(U,Q)}(\operatorname{\mathsf{\tau-tilt}}A)$.

Lemma 2.3.

[CWZ23, Corollary 3.12] Let $(M,P)$ be a basic $\tau$-tilting pair of $A$ and $(U,Q)$ a basic $\tau$-rigid pair which is a common direct summand of $(M,P)$ and $(A,0)$. Assume that there is a path from $(A,0)$ to $(M,P)$ in $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$, then there is a path from $(A,0)$ to $(M,P)$ in the subgraph $\mathcal{H}_{(U,Q)}(\operatorname{\mathsf{\tau-tilt}}A)$.

For a functorially finite torsion class $\mathcal{T}$ of $\operatorname{\mathsf{mod}}\nolimits A$, we denote by $P(\mathcal{T})$ the direct sum of one copy of each of the indecomposable $\operatorname{\mathsf{Ext}}$-projective objects in $\mathcal{T}$ up to isomorphism. In particular, $P(\mathcal{T})$ is a basic support $\tau$-tilting $A$-module. Conversely, for a basic $\tau$-rigid pair $(U,R)$, ${}^{\perp}\tau U\cap R^{\perp}$ is a functorially finite torsion class of $\operatorname{\mathsf{mod}}\nolimits A$ and $U$ is a direct summand of $P(^{\perp}\tau U\cap R^{\perp})$ (cf. [AIR14]).

Let $(U,R)$ be a basic $\tau$-rigid pair of $A$. Denote by $\operatorname{\mathsf{\tau-rigid-pair}}A$ the set of isomorphism classes of basic $\tau$-rigid pairs of $A$ and $\operatorname{\mathsf{\tau-rigid-pair}}_{(U,R)}A$ the subset of $\operatorname{\mathsf{\tau-rigid-pair}}A$ consisting of basic $\tau$-rigid pairs which admit $(U,R)$ as a direct summand. Let $T=P(^{\perp}\tau U\cap R^{\perp})$ and $B=\operatorname{\mathsf{End}}T/\langle e_{U}\rangle$, where $e_{U}$ is the idempotent of $\operatorname{\mathsf{End}}T$ associated with $U$. The following is known as the $\tau$-reduction theory of $A$ with respect to $(U,R)$ (cf. [J15, Thoerem 3.16] and [FGLZ23, Corollary A.4]).

Lemma 2.4.

Keep the notation as above. There is an order-preserving bijection

$$E_{(U,R)}:\operatorname{\mathsf{\tau-rigid-pair}}_{(U,R)}A\to\operatorname{\mathsf{\tau-rigid-pair}}B$$

which commutes with direct sums and restricts to a bijection

$$E_{(U,R)}:\operatorname{\mathsf{\tau-tilt}}_{(U,R)}A\to\operatorname{\mathsf{\tau-tilt}}B$$

commuting with mutations. In particular, there is an isomorphism of graphs between $\mathcal{H}_{(U,R)}(\operatorname{\mathsf{\tau-tilt}}A)$ and $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}B)$.

2.2. Wall and chamber structure

We now recall a construction from [BST19]. Let $A$ be a finite dimensional $k$-algebra and $\{e_{1},\dots,e_{n}\}$ a complete set of pairwise orthogonal idempotents of $A$. Let $P(i)=e_{i}A$ be the indecomposable projective $A$-module associated with $e_{i}$ and $S_{i}=\operatorname{\mathsf{top}}P(i)$ its simple top, where $1\leq i\leq n$. We identify the Grothedieck group $K_{0}(\operatorname{\mathsf{mod}}\nolimits A)$ of $\operatorname{\mathsf{mod}}\nolimits A$ with $\mathbb{Z}^{n}$ via the function

$$\underline{\dim}\,:\operatorname{\mathsf{mod}}\nolimits A\to\mathbb{Z}^{n}$$

which maps $S_{i}$ to $\mathbf{e}_{i}$, where $\mathbf{e}_{1},\dots,\mathbf{e}_{n}$ is the standard basis of $\mathbb{Z}^{n}$. Denote by $\langle-,-\rangle$ the canonical inner product of $\mathbb{R}^{n}$. For any vector $\theta\in{\mathbb{R}}^{n}$, a non-zero $A$-module $M$ is called $\theta$-semistable if $\langle\theta,\underline{\dim}\,M\rangle=0$ and $\langle\theta,\underline{\dim}\,L\rangle\leq 0$ for every submodule $L$ of $M$. The stability space of an $A$-module $M$ is then defined as

$$\mathfrak{D}_{A}(M)=\{\theta\in{\mathbb{R}}^{n}~{}|~{}\text{$M$ is $\theta$-semistable}\}.$$

We say that $\mathfrak{D}_{A}(M)$ is a wall of $A$ when $\mathfrak{D}_{A}(M)$ has codimension one.

Outside the walls, there are only vectors $\theta$ having no non-zero $\theta$-semistable modules. Removing the closure of all walls, we obtain a set

$$\mathfrak{R}_{A}={\mathbb{R}}^{n}\backslash\overline{\bigcup_{M\in\operatorname{\mathsf{mod}}\nolimits A}\mathfrak{D}_{A}(M)}$$

whose connected components $\mathfrak{C}$ are called chambers. As connected components of an open set in ${\mathbb{R}}^{n}$, the chambers have dimension $n$. This decomposition of ${\mathbb{R}}^{n}$ is called the wall and chamber structure of the algebra $A$ on ${\mathbb{R}}^{n}$.

The following is an easy observation (cf. [BST19, Lemma 4.13]).

Lemma 2.5.

Let $A$ be a finite dimensional $k$-algebra and $B$ a quotient algebra of $A$ with $|B|=|A|$. Then every wall of $B$ is also a wall of $A$.

Next, let $(M,P)$ be a $\tau$-rigid pair. Write $P$ as a direct sum of indecomposable projective $A$-modules as

$$P=\bigoplus_{i=1}^{n}P(i)^{c_{i}},$$

where $c_{1},\dots,c_{n}$ are non-negative integers. Let

$$\bigoplus_{i=1}^{n}P(i)^{b_{i}}\xrightarrow{f_{M}}\bigoplus_{i=1}^{n}P(i)^{a_{i}}\to M\to 0$$

be a minimal projective presentation of $M$, where $a_{i},b_{j}$ are non-negative integers. Recall that the $g$-vector $g_{(M,P)}$ associated with $(M,P)$ is defined as

$$g_{(M,P)}=[a_{1}-b_{1},\dots,a_{n}-b_{n}]^{t}-\sum_{i=1}^{n}c_{i}\mathbf{e}_{i}.$$

It is known that different $\tau$-rigid pairs have different $g$-vectors and the $n$ $g$-vectors of indecomposable direct summands of a basic $\tau$-tilting pair form a basis of $\mathbb{Z}^{n}$ (cf. [AIR14]).

The following fact was first noticed in [BST19] and was later shown in [A21].

Theorem 2.6.

Let $A$ be a finite-dimensional $k$-algebra. Then there is an injective function $\mathfrak{C}$ mapping the $\tau$-tilting pair $(M,P)$ onto a chamber $\mathfrak{C}_{(M,P)}$ and every chamber arises this way. Moreover, $\tau$-tilting pair $(M^{\prime},P^{\prime})$ is a mutation of $(M,P)$ if and only if $\mathfrak{C}_{(M^{\prime},P^{\prime})}$ is a neighbor of $\mathfrak{C}_{(M,P)}$, namely, they are separated by a wall.

More precisely, let $(M,P)=\bigoplus_{i=1}^{n}(M_{i},P_{i})$ be a basic $\tau$-tilting pair with indecomposable direct summands $(M_{i},P_{i}),1\leq i\leq n$. The chamber $\mathfrak{C}_{(M,P)}$ associated with $(M,P)$ is defined as

$$\mathfrak{C}_{(M,P)}=\left\{\sum_{i=1}^{n}k_{i}g_{(M_{i},P_{i})}~{}|~{}0<k_{i}\in\mathbb{R}\right\}.$$

In other words, $\mathfrak{C}_{(M,P)}$ is the interior of the positive cone

$${C_{(M,P)}}=\left\{\sum_{i=1}^{n}k_{i}g_{(M_{i},P_{i})}~{}|~{}0\leq k_{i}\in\mathbb{R}\right\}.$$

Recall that a smooth path $\gamma:[0,1]\rightarrow{\mathbb{R}}^{n}$ is a $\mathfrak{D}_{A}$-generic path if:

• $\gamma(0)$ and $\gamma(1)$ are located inside some chambers;

• If $\gamma(t)$ belongs to the intersection $\mathfrak{D}_{A}(M)\cap\mathfrak{D}_{A}(N)$ of two walls, then the dimension vector $\underline{\dim}\,M$ of $M$ is a scalar multiple of the dimension vector $\underline{\dim}\,N$ of $N$;

• whenever $\gamma(t)$ is in $\mathfrak{D}_{A}(M)$, then $\langle\gamma^{\prime}(t),\underline{\dim}\,M\rangle\neq 0.$

That is, a smooth path is $\mathfrak{D}_{A}$-generic if it crosses one wall at a time and the crossing is transversal.

Lemma 2.7.

Let $(M,P)$ and $(N,Q)$ be basic $\tau$-tilting pairs.

• If $\gamma$ is a $\mathfrak{D}_{A}$-generic path crossing finitely many walls such that $\gamma(0)\in\mathfrak{C}_{(M,P)}$ and $\gamma(1)\in\mathfrak{C}_{(N,Q)}$, then there is a finite mutation sequence from $(M,P)$ to $(N,Q)$.

• Conversely, if there is a finite mutation sequence from $(M,P)$ to $(N,Q)$, then there is a $\mathfrak{D}_{A}$-generic path $\gamma$ crossing finitely many walls such that $\gamma(0)\in\mathfrak{C}_{(M,P)}$ and $\gamma(1)\in\mathfrak{C}_{(N,Q)}$.

Proof.

Taking into account the description of chambers, the first statement is a direct consequence of Theorem 2.6. The converse statement is proved in [BST19]. ∎

3. The main results

In this section we are first interested in whether the connectivity of the $\tau$-tilting graph of an algebra is maintained under $\tau$-reductions. Then we examine quotients that are more general than $\tau$-reductions.

3.1. Reduction

Let $A$ be a finite dimensional $k$-algebra and $(U,R)$ a basic $\tau$-rigid pair of $A$. Denote by $T=P(^{\perp}\tau U\cap R^{\perp})$ and $B=\operatorname{\mathsf{End}}T/\langle e_{U}\rangle$.

Theorem 3.1.

If $A$ has reachable-in-face property, then so does $B$. Assume moreover that $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ is connected, then so is $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}B)$.

Proof.

Let $(M^{\prime},P^{\prime})$ and $(N^{\prime},Q^{\prime})$ be two basic $\tau$-tilting pairs of $B$ with a common direct summand $(L^{\prime},Z^{\prime})$. According to Lemma 2.4, we set $(M,P)=E_{(U,R)}^{-1}(M^{\prime},P^{\prime})$, $(N,Q)=E_{(U,R)}^{-1}(N^{\prime},Q^{\prime})$ and $(L,Z)=E_{(U,R)}^{-1}(L^{\prime},Z^{\prime})$. In particular, $(U,R)$ is a common direct summand of $(M,P)$, $(N,Q)$ and $(L,Z)$. Furthermore, $(L,Z)$ is a common direct summand of $(M,P)$ and $(N,Q)$.

Now assume that $A$ has reachable-in-face property and there is a path connecting $(M^{\prime},P^{\prime})$ and $(N^{\prime},Q^{\prime})$ in $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}B)$. Again by Lemma 2.4, there is a path connecting $(M,P)$ and $(N,Q)$ in $\mathcal{H}_{(U,R)}(\operatorname{\mathsf{\tau-tilt}}A)$. Since $(L,Z)$ is a common direct summand of $(M,P)$ and $(N,Q)$, there is a path connecting $(M,P)$ and $(N,Q)$ in $\mathcal{H}_{(L,Z)}(\operatorname{\mathsf{\tau-tilt}}A)$. According to Lemma 2.4, there is a path linking $(M^{\prime},P^{\prime})$ and $(N^{\prime},Q^{\prime})$ in $\mathcal{H}_{(L^{\prime},Z^{\prime})}(\operatorname{\mathsf{\tau-tilt}}B)$. Hence $B$ has reachable-in-face property. Suppose moreover that $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ is connected. It follows that $\mathcal{H}_{(U,R)}(\operatorname{\mathsf{\tau-tilt}}A)$ is connected. Consequently, $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}B)$ is connected. ∎

3.2. Quotients

Note that a quotient algebra of a basic finite dimensional $k$-algebra is always basic. We begin with the following observation.

Lemma 3.2.

Let $A$ be a finite dimensional $k$-algebra and $B$ be a quotient algebra of $A$. If $|B|<|A|$, then there is a non-zero primitive idempotent element $e$ of $A$ such that $B$ is a quotient algebra of $A/\langle e\rangle$.

Proof.

Note that $B$ is a quotient algebra of $A$, there is a surjective homomorphism $f:A\rightarrow B$. Suppose that $|A|=n$ and $\{e_{1},e_{2},\dots,e_{n}\}$ is a complete set of primitive pairwise orthogonal idempotent elements with $1_{A}=e_{1}+e_{2}+\cdots+e_{n}$. For each $1\leq i\leq n$, set $f_{i}=f(e_{i})$, then $1_{B}=f(1_{A})=f_{1}+f_{2}+\cdots+f_{n}$. If $f_{i}\neq 0$ for each $1\leq i\leq n$, then $\{f_{1},f_{2},\dots,f_{n}\}$ is a complete set of primitive pairwise orthogonal idempotent elements of $B$, therefore $|B|=n=|A|$, which contradicts with the assumption that $|B|<|A|$. So there is some $i$ such that $f_{i}=0$, i.e. $f(e_{i})=0$. Let $\pi_{i}:A\rightarrow A/\langle e_{i}\rangle$ be the canonical epimorphism, there exists an epimorphism $g$ from $A/\langle e_{i}\rangle$ to $B$ such that $g\pi_{i}=f$. Hence $B$ is a quotient algebra of $A/\langle e_{i}\rangle$. ∎

Proposition 3.3.

Let $A$ be a finite dimensional $k$-algebra and $e$ a primitive idempotent element of $A$. If the $\tau$-tilting graph of $A$ is connected, then the $\tau$-tilting graph of $A/\langle e\rangle$ is also connected.

Proof.

Note that $eA$ is a projective $A$-module and $P(^{\perp}\tau eA)=A$. Applying Lemma 2.4 to the basic $\tau$-rigid pair $(eA,0)$, we have an order-preserving bijection

$$E_{(eA,0)}:\operatorname{\mathsf{\tau-tilt}}_{(eA,0)}A\to\operatorname{\mathsf{\tau-tilt}}A/\langle e\rangle.$$

Let $(M^{\prime},P^{\prime})$ and $(N^{\prime},Q^{\prime})$ be two basic $\tau$-tilting pairs of $A/\langle e\rangle$. Set $(M,P):=E_{(eA,0)}^{-1}(M^{\prime},P^{\prime})$ and $(N,Q):=E_{(eA,0)}^{-1}(N^{\prime},Q^{\prime})$.

Since the $\tau$-tilting graph of $A$ is connected, there is a path connecting $(M,P)$ and $(A,0)$ in $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$. Note that $(eA,0)$ is a common direct summand of $(M,P)$ and $(A,0)$, there is a path in $\mathcal{H}_{(eA,0)}(\operatorname{\mathsf{\tau-tilt}}A)$ connecting $(M,P)$ and $(A,0)$ by Lemma 2.3. Similarly, there is a path in $\mathcal{H}_{(eA,0)}(\operatorname{\mathsf{\tau-tilt}}A)$ connecting $(N,Q)$ and $(A,0)$. We conclude that there is path connecting $(M^{\prime},P^{\prime})$ and $(N^{\prime},Q^{\prime})$ in $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A/\langle e\rangle)$ by Lemma 2.4 ∎

Proposition 3.4.

Let $A$ be a finite dimensional $k$-algebra and $B$ a quotient algebra of $A$ with $|A|=|B|$. Suppose each $B$-chamber contains at least one $A$-chamber. If the $\tau$-tilting graph of $A$ is connected, then the $\tau$-tilting graph of $B$ is also connected.

Proof.

Let $(M,P)$ and $(N,Q)$ be two basic $\tau$-tilting pairs of $B$. According to Lemma 2.6, there are two corresponding chambers $\mathfrak{C}^{B}_{(M,P)}$ and $\mathfrak{C}^{B}_{(N,Q)}$ in $\mathfrak{R}_{B}$. By Lemma 2.5, each $B$-wall is also an $A$-wall. By the assumption that each $B$-chamber contains at least one $A$-chamber, there are chambers $\mathcal{C}_{1}$ and $\mathcal{C}_{2}$ in $\mathfrak{R}_{A}$ such that $\mathcal{C}_{1}\subset\mathfrak{C}^{B}_{(M,P)}$ and $\mathfrak{C}_{2}\subset\mathfrak{C}^{B}_{(N,Q)}$. Note that the $\tau$-tilting graph of $A$ is connected, there is a $\mathfrak{D}_{A}$-generic path $\gamma$ with $\gamma(0)\in\mathfrak{C}_{1}$ and $\gamma(1)\in\mathfrak{C}_{2}$ by Lemma 2.7. Since each $B$-wall is an $A$-wall, $\gamma$ is also a $\mathfrak{D}_{B}$-generic path. Note that $\gamma(0)\in\mathfrak{C}_{1}\subset\mathfrak{C}^{B}_{(M,P)}$ and $\gamma(1)\in\mathfrak{C}_{2}\subset\mathfrak{C}^{B}_{(N,Q)}$, by Lemma 2.7 again, we conclude that there is a path connecting $(M,P)$ and $(N,Q)$ in $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}B)$. ∎

3.3. Application to $g$-tame algebras

Let $A$ be a finite dimensional $k$-algebra with $|A|=n$. Recall from [AK23, Definition 7.6] that $A$ is $g$-tame if $\overline{\mathcal{F}(A)}={\mathbb{R}}^{n}$, where

$$\mathcal{F}(A)=\bigcup_{(M,P)\in\operatorname{\mathsf{\tau-tilt}}A}C_{(M,P)}.$$

We need the following property of $g$-tame algebras (cf. [PYK23, Proposition 3.11] and [AK23, Corrolary 7.8]).

Lemma 3.5.

Let $A$ be a finite dimensional $k$-algebra and $B$ a quotient algebra of $A$. If $A$ is $g$-tame, then so is $B$.

We are now ready to state and prove the main result of this section.

Theorem 3.6.

Let $A$ be a $g$-tame algebra and $B$ a quotient algebra of $A$. If the $\tau$-tilting graph $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ of $A$ is connected, then so is $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}B)$.

Proof.

If $|A|=|B|$, then each $B$-wall is also an $A$-wall by Lemma 2.5. Since $A$ is $g$-tame, each $B$-chamber contains at least one $A$-chamber. Then we are done by Proposition 3.4.

So assume now that $|A|>|B|$. By Lemma 3.2, there is a non-zero idempotent element $e$ of $A$ such that $B$ is a quotient of $A/\langle e\rangle$ with $|B|=|A/\langle e\rangle|$. According to Proposition 3.3, the $\tau$-tilting graph of $A/\langle e\rangle$ is also connected. On the other hand, $A/\langle e\rangle$ is also $g$-tame by Lemma 3.5. Now the result follows from the case $|A|=|B|$. ∎

It is known that the $\tau$-tilting graph $\mathcal{H}(\operatorname{\mathsf{\tau-tilt}}A)$ is connected if $A$ belongs to one of the following classes of algebras:

• cluster-tilted algebras arising from hereditary abelian categories [BMRRT, BKL10, FG21];

• $2$-Calabi-Yau tilted algebras arising from marked surfaces except closed surfaces with exactly one puncture [QZ17, Y20];

• gentle algebras [FGLZ23];

• skew-gentle algebras [HZZ22].

Note that $2$-Calabi-Yau tilted algebra arising from marked surfaces and skew-gentle algebras are tame algebras. According to [PYK23], tame algebras are $g$-tame. As a consequence of Theorem 3.6, we obtain a large class of algebras with connected $\tau$-tilting graphs.

Corollary 3.7.

Let $A$ be a skew-gentle algebra or a cluster-tilted algebra of tame type, or a $2$-Calabi-Yau tilted algebra arising from a marked surface that is not closed with exactly one puncture. Let $B$ be any quotient algebra of $A$. Then the $\tau$-tilting graph of $B$ is connected.

It is known quasi-tilted algebras are quotients of cluster-tilted algebras arising from hereditary abelian categories (cf. [ABS08, ASS17, Z06]), then by Corollary 3.7, we have,

Corollary 3.8.

Let $A$ be a quotient algebra of a quasi-tilted algebra of tame type. Then the $\tau$-tilting graph of $A$ is connected.